Skip to content
AtheronLABS

You're visiting from the United States. Prices are shown in US dollars. Not right?

labs@atheron:~/industries/staffing/ai$ train agent --grounded

AI for recruiters

Candidate matching you can explain.

Hiring is where AI can do most harm if it is careless. A matching tool should find people a keyword search misses, show why it ranked them and never reject anyone on its own. We run it on an open-source model in your cloud, because résumés are personal information, and test it for fairness before it is used.

// what matters here

What is different in this industry.

01

The recruiter decides

The tool ranks and explains. Who is contacted, interviewed or placed is a recruiter's decision, and the system records who made it.

02

Bias tested, not assumed

Names, photos and other details that could skew a ranking are hidden from the model, and outcomes are compared across groups before launch and regularly after.

03

Candidate data stays with you

Résumés are processed by an open-source model in your own cloud account. Nothing is sent to an outside model provider.

04

Transparent to candidates

Where rules require it, such as provinces that ask employers to disclose AI in hiring, the tool supports the notices and records you need.

// example projects

Priced examples for this service.

Each one is hypothetical, labelled as such, and priced live from our rate card. Open any of them in the estimator and make it yours.

Example projectWeb appAIOpen-source model

Recruiter-led candidate matching

Hypothetical. Not a client, not a result.

// The problem

Recruiters search the candidate database by keyword and miss people who have the right experience described in other words.

// What we would build

A matching tool on an open-source model served in the agency's own AWS account, chosen because résumés are personal information that should not go to an outside provider, that ranks candidates for an order with the reasons beside each one. The recruiter decides who to contact, and every ranking is tested for unfair patterns before launch.

// What is in it

  • Candidates matched on skills, certifications, experience and availability, with the reasons shown
  • Names, photos and other details that could bias a ranking hidden from the model
  • The recruiter chooses who to contact; the tool never rejects anyone
  • Outcomes tested across groups before launch and on a schedule after
  • An open-source model in your own cloud account, with no candidate data sent outside

// Stack

  • Qwen or Llama (open-source)
  • vLLM
  • PostgreSQL with pgvector
  • Applicant tracking system integration
  • AWS

// estimate

Build
≈ US$108,000 to US$166,000, delivered within 21 weeksCAD 154,400 to 235,900
Hosting
≈ US$6,710 a monthCAD 9,550 a month
Support
≈ US$2,500 a monthCAD 3,565 a month

Prices in your currency are estimates from today's Bank of Canada rate. All invoicing is in CAD or USD.

AI route
Open-source model
Model running cost
≈ US$6,530 a monthCAD 9,300 a month

Timeline by milestone

Discovery
2.9 to 3.4 weeks
Specification and evaluation plan
0.4 weeks
Design approved
2.4 to 3.9 weeks
Core features
1.4 to 1.9 weeks
Full build
0.9 to 1.4 weeks
Working pilot
3.4 to 5.4 weeks
Testing and fixes
0.9 to 1.9 weeks
Launch
0.9 to 1.4 weeks
Production
0.9 to 1.4 weeks

Outside our hands, and added to the calendar

API access from your vendor
1 to 4 weeks

How it is paid

Deposit 20%
CAD 30,820 to 47,120
Discovery 2%
CAD 3,082 to 4,712
Specification and evaluation plan 10.3%
CAD 15,872.30 to 24,266.80
Design approved 2%
CAD 3,082 to 4,712
Core features 6%
CAD 9,246 to 14,136
Full build 4%
CAD 6,164 to 9,424
Working pilot 31.1%
CAD 47,925.10 to 73,271.60
Testing and fixes 2%
CAD 3,082 to 4,712
Launch 2%
CAD 3,082 to 4,712
Production 10.6%
CAD 16,334.60 to 24,973.60
Holdback, 30 days after launch (10%)
CAD 15,410 to 23,560
GPU time for training and testing, at cost
CAD 300. Billed up front, at cost, outside the milestones

// questions

Questions we get about this.

Is it legal to use AI to screen job candidates?

It depends on where you hire and how the tool is used, and some places now require disclosure or assessments. Your advisers decide; we build the tool to the requirements they set and keep the decision with a person.

Will it reject applicants automatically?

No. It ranks and explains; it never removes anyone from consideration. Recruiters see the whole pool and can search it however they like.

How do you test a matching model for bias?

We compare rankings and outcomes across groups, using test data built for the purpose, before launch and on a schedule after. Results go to you in writing, and a failing test stops a release.

// next

Not quite your project?

Tell us what you have in mind. We will come back to you with a range and the questions that would narrow it. Or book a call and talk it through.

AI candidate matching for staffing agencies and recruiters | Atheron Network Labs